Richard M. Bailey

Although mechanisms forming the various dune types are well established, the drivers of morphological variability within linear dunes remain understudied. To address this, we systematically mapped key dune morphometrics along with potential controlling factors in the linear dune fields of the south‐western Kalahari and statistically assessed their relationships. Mapping techniques are based on digital terrain analysis and also include a novel application of deep learning to predict wavelengths. Results highlight the influence of wind regime and river valleys on dune morphology and suggest that the south‐western Kalahari reflects regionally heterogeneous controls. The study demonstrates how data‐driven methods suited for the analysis across large spatial scales can reveal dune–environment relationships and identify key patterns and anomalies that warrant further research.

So-called `wicked problems', those involving complex multi-dimensional settings, non-verifiable outcomes, heterogeneous impacts and a lack of single objectively correct answers, have plagued humans throughout history. Modern examples include decisions over justice frameworks, solving environmental pollution, planning for pandemic resilience and food security. The use of state-of-the-art artificial intelligence systems (notably Large Language Model-based agents) collaborating with humans on solving such problems is being actively explored. While the abilities of LLMs can be improved by, for example, fine-tuning, hand-crafted system prompts and scaffolding with external tools, LLMs lack endogenous mechanisms to develop expertise through experience in such settings. This work address this gap with Dialectica, a framework where agents engage in structured dialogue on defined topics, augmented by memory, self-reflection, and policy-constrained context editing. Formally, discussion is viewed as an implicit meta-reinforcement learning process. The `dialogue-trained' agents are evaluated post-hoc using judged pairwise comparisons of elicited responses. Across two model architectures (locally run Qwen3:30b and OpenAI's o4-mini) results show that enabling reflection-based context editing during discussion produces agents which dominate their baseline counterparts on Elo scores, normalized Bradley-Terry-Davidson ability, and AlphaRank mass. The predicted signatures of learning are observed qualitatively in statement and reflection logs, where reflections identify weaknesses and reliably shape subsequent statements. Agreement between quantitative and qualitative evidence supports dialogue-driven context evolution as a practical path to targeted expertise amplification in open non-verifiable domains.

Democracies around the world are facing unprecedented and interconnected challenges, including polarisation, misinformation, and democratic backsliding. These challenges are in part informational in nature, prompting researchers from disparate fields of inquiry to study how the consumption and sharing of information impacts relevant attitudes, beliefs, and behaviours. However, information flows through complex and dynamic ecosystems that play out as an interaction between media outlets, social media channels, and socio-political contexts – all filtered by individual psychology. We argue that it is time to move away from the siloed study of informational challenges, and to establish ‘infodemiology’ as a distinct interdisciplinary field, dedicated to understanding how information flows and beliefs may form within these intricate systems. We argue that studying information ecosystems dynamically rather than in isolation yields critical insights that exceed the sum of their individual components. We argue that a minimal infodemiological model must at least four component categories (citizens, broadcast media, social networks, and a regulatory system). By considering the interplay between individual-level (e.g., trust in institutions, susceptibility to manipulation) and system-level (e.g., recommender algorithms, network structures) components of information environments, we argue that the establishment of infodemiology as a distinct field would encourage theoretical and methodological innovations, spark discussions on how to calibrate and validate infodemiological models, and invite practical benefits such as policy interventions.

Unambiguous identification of the rewards driving behaviours of entities operating in complex open-ended real-world environments is difficult, partly because goals and associated behaviours emerge endogenously and are dynamically updated as environments change. Reproducing such dynamics in models would be useful in many domains, particularly where fixed reward functions limit the adaptive capabilities of agents. Simulation experiments described assess a candidate algorithm for the dynamic updating of rewards, RULE: Reward Updating through Learning and Expectation. The approach is tested in a simplified ecosystem-like setting where experiments challenge entities' survival, calling for significant behavioural change. The population of entities successfully demonstrate the abandonment of an initially rewarded but ultimately detrimental behaviour, amplification of beneficial behaviour, and appropriate responses to novel items added to their environment. These adjustment happen through endogenous modification of the entities' underlying reward function, during continuous learning, without external intervention.

The linear production and consumption of plastics today is unsustainable. It creates large amounts of unnecessary and mismanaged waste, pollution and carbon dioxide emissions, undermining global climate targets and the Sustainable Development Goals. This Perspective provides an integrated technological, economic and legal view on how to deliver a circular carbon and plastics economy that minimizes carbon dioxide emissions. Different pathways that maximize recirculation of carbon (dioxide) between plastics waste and feedstocks are outlined, including mechanical, chemical and biological recycling, and those involving the use of biomass and carbon dioxide. Four future scenarios are described, only one of which achieves sufficient greenhouse gas savings in line with global climate targets. Such a bold system change requires 50% reduction in future plastic demand, complete phase-out of fossil-derived plastics, 95% recycling rates of retrievable plastics and use of renewable energy. It is hard to overstate the challenge of achieving this goal. We therefore present a roadmap outlining the scale and timing of the economic and legal interventions that could possibly support this. Assessing the service lifespan and recoverability of plastic products, along with considerations of sufficiency and smart design, can moreover provide design principles to guide future manufacturing, use and disposal of plastics.

To effectively manage complex human-environment fisheries systems, it is necessary to understand the psychology of fisher agents. While bio-economic models typically provide simple, abstract approaches for human behaviour (e.g., fully informed profit maximisers), fisher agents are of course neither simple nor perfect. Imperfections of learning, memory, and information availability, combined with the diversity of value preferences within populations, can lead to substantial deviations and unanticipated effects of interventions. This paper presents a computational model of fisher agents' decision-making that draws on theoretical and empirical psychological insights to enrich this critical component. The model includes mechanisms for information integration (learning), social comparisons, and thresholds for economic satisfaction. In offering this enriched account, the model captures how fishers may adapt behaviourally given changes in policy, economic conditions, or social pressures. Furthermore, the model can be parameterised to capture the effects of different sociocultural contexts can be simulated. The model of fisher agents has been implemented as part of POSEIDON (an agent-based fisheries management model), showing that fishers imbued with the model learn and adapt when responding dynamically to changing conditions. The model is thus demonstrated in a fisheries environment, but we discuss how its architecture could be implemented for simulation in other human-environment systems, such as designing policies to combat the human-environment problems.

Agricultural transformation is imperative to achieve environmental and social objectives. However, the complexity and variety of agricultural systems, and the myriad factors impacting farmer decision-making often hinder policy efficacy. Our research provides a new, integrated perspective on this issue, using a model that incorporates the complexities of both farmers and farm systems to enable more complete explanatory mechanisms for policy success and failure. We unpack a hierarchy of barriers to change, noting how those barriers interact, including exposition of previously ill-understood factors (e.g., farmer psychology). Via inclusion of social and complex system dynamics, we also show how policy bundles can exploit tipping-point dynamics. Our findings provide an explanation for disparate agricultural system outcomes, demonstrating the crucial role of farmer psychology and system-level dynamics in the success or failure of regenerative policies. This underscores the need for tailored, multi-faceted policy approaches in promoting sustainable agriculture.

It is important to incorporate fisher motivations and behavior into fisheries management models. Incorrect behavioral assumptions may yield ineffective incentives or interventions or even produce unintended consequences. To understand fisher behavior in a developing country, we surveyed 93 Indonesian snapper fishers. Results suggest they consider competing aspects such as income, personal reputation, and sociocultural norms when deciding where and what to fish; they update beliefs about location, bountifulness, and catchability of target fish stocks through direct observations, inferences over geographical similarities, and social interactions with other fishers, and they evaluate satisfaction economically as well as socially. Information sharing and social knowledge are likely port-specific, representing local sociocultural norms rather than being related to vessel size, target catch, or other demographics. The prevalence of information sharing and imitation patterns suggests that fisher decision-making in Indonesian snapper fisheries has a significant sociocultural component. We discuss implications for fisheries management models and for policy decisions.

The Pleistocene presence of the genusHomoin continental Southeast Asia is primarily evidenced by a sparse stone tool record and rare human remains. Here we report a Middle Pleistocene hominin specimen from Laos, with the discovery of a molar from the Tam Ngu Hao 2 (Cobra Cave) limestone cave in the Annamite Mountains. The age of the fossil-bearing breccia ranges between 164–131 kyr, based on the Bayesian modelling of luminescence dating of the sedimentary matrix from which it was recovered, U-series dating of an overlying flowstone, and U-series–ESR dating of associated faunal teeth. Analyses of the internal structure of the molar in tandem with palaeoproteomic analyses of the enamel indicate that the tooth derives from a young, likely female,Homoindividual. The close morphological affinities with the Xiahe specimen from China indicate that they belong to the same taxon and that Tam Ngu Hao 2 most likely represents a Denisovan.